Fast Codebook Design for Vector Quantization on Partitioned Space

Vector quantization(VQ) is an attractive technique for lossy data compression, which has been a key technology for data storage and/or transfer. So far, various competitive learning (CL) algorithms have been proposed to design optimal codebooks presenting quantization with minimized errors. However, their practical use has been limited for large scale problems, due to the computational complexity of competitive learning. This paper presents a fast codebook design algorithm reducing the computational complexity. The algorithm partitions input-vector space into some subspaces, and designs a subcodebook for each subspace independently. Experimental results indicate a speedup of this algorithm without serious increase of quantization errors, and promising potential of its parallelization.